Do not spoil the dataviz dish

The technical side of data visualization is basically solved.

Last Monday I gave a live demo. In 8 minutes, in front of 300 people, I asked the audience for their favorite tool and their biggest frustration. I grabbed every message from the chat, turned them into a dataset, and built 2 interconnected interactive charts. Live.

All it takes is the right AI setup, some foundational knowledge, a few magic ingredients, and a clear description of the chart of your dreams.

Check the demo

(By the way, that's exactly what I teach here.)

But this superpower can backfire, and I'm seeing it happen every single day now. So today, I would like to share an important warning.

⚠️ Don't spoil the dish

With such a powerful tool in hand, it's very tempting to build crazy, complicated, fully furnished visual pages.

Something like this:

And it's already happening!

I recently judged 30 entries for a dataviz challenge. All of them used the D3 + React stack. Most were technically stunning. But most were really hard to follow.

One of my favorite sources of inspiration, Josh Comeau, has a line about animation on the web that I love:

"Animation is like salt: too much of it spoils the dish."

It's true for animation. But honestly, it's true for tech wizardry in general and it's becoming the number one issue I see.

💡 What to do instead

I believe it's important to question every single item of the page you're building. Examples:

🔵 You built a fancy dataset transition that lasts 2 seconds every time someone clicks?
--> Make it 0.5 seconds. And if it doesn't help the reader follow a group or a trend, remove it.

🔵 Added a button to switch from a box plot to a violin plot?
--> If both tell the exact same story, that's just noise.

🔵 Wrote a long paragraph explaining how the chart and the app work?
--> Question your design. People don't read. Everything should be self-explanatory.

🔵 Used 10 datasets to share 20 facts about your topic?
--> People will get lost. Find one story only and tell it well.

🔵 Added a tooltip to your barplot that just shows the bar value?
--> Probably useless.

🔵 Created interactive particles floating all over your text?
--> Distracting. Remove them!

Tech for tech's sake barely impresses anyone anymore.

What makes people listen to your story is something simple, well thought out, and crystal clear, with no distractions.

The bottom line

With AI, we're drowning in content, and data visualization is no exception.

Your audience has very limited time for your project. They want to understand your main point, not a zillion things. Their attention is precious, and they'll give up very easily.

So go straight to the point. Use interactivity, animation, buttons, and text with care, only to serve your main story and nothing else.

Less is more has never been so true.

Best,
Yan